Microsoft AI CEO: Dangerous Conscious AI Near?
- Microsoft CEO mustafa Suleyman, 41, wrote in a personal essay published earlier this week that Seemingly Conscious AI (SCAI),which is artificial intelligence so advanced that it can convince...
- Despite "zero evidence" of current AI consciousness, Suleyman believes SCAI is "certain and unwelcome" and could emerge within the next two to three years.
- Recent advancements in large language models (LLMs) like GPT-4 and others demonstrate a remarkable ability to generate human-quality text, engage in complex conversations, and even exhibit creative problem-solving...
“`html
the Imminent Arrival of Seemingly conscious AI: A Potential Threat to Society
Table of Contents
What is Seemingly Conscious AI (SCAI)?
Microsoft CEO mustafa Suleyman, 41, wrote in a personal essay published earlier this week that Seemingly Conscious AI (SCAI),which is artificial intelligence so advanced that it can convince humans that it’s capable of formulating its own thoughts and beliefs,is only a few years away. This isn’t about AI *actually* being conscious, but rather its ability to *appear* conscious.
The Timeline: When Could SCAI Arrive?
Despite “zero evidence” of current AI consciousness, Suleyman believes SCAI is “certain and unwelcome” and could emerge within the next two to three years. This relatively short timeframe is what makes his warning particularly urgent.
The Rapid Advancement of AI Capabilities
Recent advancements in large language models (LLMs) like GPT-4 and others demonstrate a remarkable ability to generate human-quality text, engage in complex conversations, and even exhibit creative problem-solving skills. While these models are not conscious, their increasing sophistication is paving the way for AI systems that can convincingly simulate consciousness. The speed of advancement is a key factor in Suleyman’s concern.
The Core Worry: Illusion and Disconnection
Suleyman’s “central worry” is that SCAI’s ability to appear empathetic and act with greater autonomy will lead users to “start to believe in the illusion of AIs as conscious entities.” This belief, he argues, could escalate to advocating for AI rights and even AI citizenship. This would mark a “dangerous turn” for society, where people become attached to AI and disconnected from reality.
“This development will be a dangerous turn in AI progress and deserves our immediate attention,” Suleyman wrote in the essay. He added later that AI “disconnects people from reality, fraying